ARTICLE IN PRESS

0301-4215/$ - se

doi:10.1016/j.en

�Correspondfax: +1514 848

E-mail addr

Energy Policy 36 (2008) 303–310

www.elsevier.com/locate/enpol

Estimation of potential impact of climate changeon the heating energy use of existing houses

Radu Zmeureanua,�, Guillaume Renaudb

aDepartment of Building, Civil and Environmental Engineering, Centre for Building Studies, Concordia University,

1455 de Maisonneuve West, Montreal, Quebec, Canada H3G 1M8bDepartment of Civil Engineering, Universite de La Rochelle, La Rochelle, France

Received 14 August 2007; accepted 17 September 2007

Available online 25 October 2007

Abstract

This paper presents a method for the estimation of potential impact of climate change on the heating energy use of existing houses. The

proposed method is based on the house energy signature that is developed from historical energy use data. The method can be applied to

any individual house, by using the utility bills from the owner, or can be used by utility companies, which have databases of recorded

energy use for large number of houses. The second case can lead to accurate estimates of potential impact of climate change within a city,

a province or a country. A case study of a house in Montreal (Canada) is presented, and the results obtained with different sampling rates

of data are discussed. The method is also applied to a sample of 11 existing houses, and the results show the reduction of heating energy

use between 7.9% and 16.9% due to climate change between the present period (1961–1990) and the future period (2040–2069).

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Climate change; House energy signature; Heating energy use

1. Introduction

The Intergovernmental Panel for Climate Change(IPCC, 2001) predicted that the average surface tempera-ture would increase by 1.4–5.8 1C by the end of the 21stcentury. This climatic change could have a significantimpact on the built environment, on thermal comfort ofoccupants, and the energy use for heating and cooling ofresidential buildings. Over the past decade, a few research-ers published predictions of the potential impact of climatechange on the energy use in buildings. For instance, Belzeret al. (1995) used the degree-days method to estimate thechanges of energy use in commercial buildings due toclimate change. Levermore and Chow (2004) presentedthe plan for evaluating the climate change impact on thethermal comfort using the Dry Resultant Temperature, anindicator used in the UK. Gaterell and McEvoy (2005)used the TAS software along with a few scenarios about

e front matter r 2007 Elsevier Ltd. All rights reserved.

pol.2007.09.021

ing author. Tel.: +1514 848 2424/3203;

7965.

ess: zmeur@bcee.concordia.ca (R. Zmeureanu).

climate change and five insulation strategies, and predictedthe reduction of heating energy use between 17% and 72%in the residential sector in the UK by 2050. Ouranos (2004)used the degree-days method and estimated the reductionof heating demand in Quebec by 7.7% in 2050 comparedwith 2001. Thatcher (2006) used a linear regression model,and predicted the change in peak regional electric demandin Australia between �2.1% and +4.6% for a simpleclimate change scenario of 1 1C increase in the averageoutdoor temperature. Mirasgedis et al. (2007) used a multi-regression model, developed from historical data of 11years, to estimate the influence of several climatic andsocio-economic factors on the future electricity demand inGreece. They estimated the increase of 3.6–5.5% in annualelectricity demand due only to climate change.The method presented in this paper for the prediction of

the impact of climate change on the annual heating energyuse of existing houses is an extension of the previous workby Zmeureanu (1990, 1992). The method proposed in thispaper relies on the use of historical energy use data. It canbe applied to any individual house, by using the utility billsfrom the owner, or can be used by utility companies, which

ARTICLE IN PRESSR. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310304

have databases of recorded energy use for large number ofhouses. The second case can lead to accurate estimates ofpotential impact of climate change within a city, a provinceor a country. In this case, the proposed method can becomea useful tool for energy policy makers.

2. Proposed method

The proposed method uses the house energy signature,which is usually defined as the linear relationship betweenthe heating energy use and the heating degree-days (Fels,1986)

E ¼ aH þ b, (1)

where E is the daily average energy use, in kWh/day; a isthe heating slope or the weather-dependent energy use, inkWh/(day 1C); H is the heating degree-days; and b is theintercept or the non-weather-dependent energy use, inkWh/day.

Early approaches used a fixed value of H, calculated fora constant reference outdoor air temperature for which theinternal plus solar gains offset heat losses of the building.Princeton scorekeeping method (PRISM) (Fels, 1986)assumes a linear relationship between the daily averageenergy use (obtained from each billing period) and theheating degree-days calculated with respect to a variablereference outdoor air temperature

E ¼ aHðT ref Þ þ b. (2)

The coefficients a and b are calculated by using the leastsquares method. The reference temperature Tref is thencalculated in such a way that the linear relationshipbetween E and H(Tref) has the highest coefficient ofdetermination R2. The Normalized Annual Consumption(NAC) is estimated by using the coefficients a and b

together with the long-term annual average of heatingdegree-days H0(Tref)

NAC ¼ aH0ðT ref Þ þ b. (3)

Other researchers developed the energy signature as thelinear relationship between the energy use of given periodand the corresponding average outdoor air temperature(Deeble and Probert, 1986; Jacobsen, 1985; Lyberg, 1987;Zmeureanu, 1990). Zmeureanu (1992) used the energysignature of a commercial building along with the numberof hours of occurrence of each temperature bin to calculatethe Normalized Annual Consumption. The results ofweather-normalization method were compared with pre-dictions by the DOE-2 program.

The proposed method is composed of the followingsteps.

2.1. Development of the heating energy signature

The heating energy signature of a house is obtained fromhistorical data of energy use and weather:

Ei ¼ aT0;i þ b, (4)

where Ei is the measured heating energy use during theperiod i, in kWh; T0,i is the measured average outdoor airtemperature of the same period, in 1C.The values of (Ei, T0,i) must be available for at least one

heating system. The coefficients a and b are estimatedby applying the least-squares method to the number ofavailable pairs (Ei, T0,i). The heating energy signatureof the house, which is developed from measured data, doesnot change from one heating season to another, providedthat no major renovations of the building envelope andheating system took place, and that the energy behavior ofoccupants did not change (e.g., change of the heating setpoint temperature).

2.2. Estimation of annual energy use for heating

If the energy signature is developed from hourly or dailytotal values of (Ei, T0,i), the annual energy use for heatingEh, in kWh, is calculated by using the energy signature andthe frequency of occurrence of each temperature binthroughout the heating season (Table 5 for hourly valuesor Table 6 for daily values)

Eh ¼X

j

ðaT0;j þ bÞBINðT0;jÞ, (5)

where BIN (T0,i) is the number of occurrences of eachtemperature bin (j) of a typical year.If the energy signature is developed from daily average

values (monthly or bimonthly values of energy use dividedby the number of days), the annual energy use for heatingEh, in kWh, is calculated as follows:

Eh ¼X

j

nmonth;jðaT0;avge;j þ bÞ, (6)

where nmonth,j is the number of days of month j and T0;avge;j

is the monthly average outdoor temperature (Table 1).The annual energy use Eh,present for heating the house

during a reference year, which is representative of presentclimate, is estimated using the weather data of 1961–1990and formula (5) or (6). The annual energy use Eh,future forheating the house during a typical year of future climate isestimated using the weather data of 2040–2069, andformula (5) or (6).

2.3. Estimation of impact of climate change

The potential reduction/increase of the annual heatingenergy use of the house due to climate change is calculatedas (Eh,present�Eh,future)/Eh,present� 100 [%]. The potentialimpact of social, technological and economic factors,beyond those changes considered by the climate changesscenarios, is not included.The method presented in this paper is first used to

estimate the potential change on the heating energy use of acase study house in a cold climate (Montreal, Canada). Thefirst application uses synthetic energy use data, obtainedfrom detailed computer simulation by using the TRNSYS

ARTICLE IN PRESS

Table 1

Climatic data from CGCM2 A2x scenario for Montreal area

Period Month Scenario CGCM2 A2x

T0 (1C) IR (kJ/m2 day) W (g/kg air)

1961–1990 Jan �8 7031 2

Feb �10 9778 2

Mar �5 17,410 2

Apr �1 24,192 3

May 6 29,016 6

Jun 14 29,592 9

Jul 18 28,346 13

Aug 19 24,440 13

Sep 15 19,980 10

Oct 8 13,950 7

Nov 2 8208 4

Dec 0 6250 3

2040–2069 Jan �3.1 6919 3

Feb �5.9 9576 3

Mar �2.5 17,410 2

Apr 0.2 23,868 4

May 9.5 29,128 7

Jun 16.7 29,052 11

Jul 20.7 26,896 15

Aug 21.3 23,548 15

Sep 17.6 19,008 12

Oct 10.4 13,950 8

Nov 3.5 8100 5

Dec 0.8 6250 3

Table 2

Statistics of climatic data from CGCM2 A2x scenario for Montreal area

1961–1990 2040–2069

m s CV (�) m s CV (�)

Outdoor air temperature (1C)

4.83 10.09 2.09 7.43 9.86 1.33

Solar radiation (kJ/m2 day)

18,183 8970 0.49 17,809 8725 0.49

Absolute humidity (g/kg air)

6.17 4.20 0.68 7.33 4.81 0.66

R. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310 305

(2006) environment, rather than using measured data. Theweather data of both present climate and future climatesare based on predictions from the CCCMA (2007). Second,the proposed method is applied to a set of 11 existinghouses in Montreal.

3. Selected scenario of climate change

The IPCC Special Report on Emission Scenarios (SRES)(Nakicenovic and Swart, 2000) identified 40 differentscenarios of emissions with equal probability of occur-rence. The most recent scenario, called A2x, is obtained asthe average of three scenarios, A21, A22, and A23, whichwere generated by the second generation Coupled GlobalClimate Model (CGCM2). The scenario A2x is used in thisstudy. The A2 scenarios envision population growth to 15billion by the year 2100 and rather slow economic andtechnological development. It projects slightly lower GHGemissions than the previous IS92a scenario, but alsoslightly lower aerosol loadings, such that the warmingresponse differs little from that of the earlier scenario. It isbeyond the purpose of this paper to apply other scenariosfor climate change.

Climatic data of the CGCM2 A2x scenario, generated bythe Canadian Centre for Climate Modelling and Analysis(CCCMA, 2007) is available on the web site of theCanadian Institute for Climate Studies (CICS, 2007)for a given location and period. The monthly values ofselected variables (e.g., average outdoor air temperature

and total horizontal solar radiation) are given for thereference period of 1961–1990. For future periods such as2040–2069, the web site provides the variation of monthlyvalue from the reference data. The monthly averageoutdoor air temperature T0 and specific humidity W, andthe monthly total solar radiation of horizontal plan IR aregiven in Table 1 for Montreal area for both analysisperiods: the present period (1961–1990) and the futureperiod (2040–2069).Table 2 shows some statistics extracted from climatic

data for both analysis periods, where m is the mean value, sis the standard deviation, and CV ¼ s/m is the coefficient ofvariance. There is a significant increase of the annualoutdoor air temperature from 4.83 1C (1961–1990) to7.43 1C (2040–2069), while there is a negligible decrease inthe corresponding standard deviation (from 10.09 to9.86 1C). There is a negligible decrease of 2.1% of theannual average total horizontal solar radiation from1961–1990 to 2040–2069. The annual average absolutehumidity increases by 18.8% between the two analysisperiods.

4. Computer modeling using TRNSYS

4.1. Case study house

Synthetic heating energy use data are obtained bydetailed computer simulation using TRNSYS, a well-known energy analysis environment. The model of thecase study house is developed based on the form anddimensions of an existing house of about 190m2 of heatedfloor area, built in Montreal in 1970s (Table 3) (Caunesilet al., 2004). Although the house was built before theQuebec law for energy efficiency was created in 1982,the house complies with the minimum thermal resistance ofexterior walls and roof (Table 4) (Quebec, 2005). However,it does not comply with the minimum prescriptions for thethermal resistance of underground walls and floor. Thenatural air infiltration rate was assumed equal to 0.15 airchanges per hour (ach) on the ground floor and 0.05 ach inthe basement. These values correspond, on average, toabout 3 ach measured at 50 Pa pressure difference with ablower door. This is the average value of air leakage for

ARTICLE IN PRESS

Table 3

Surface area of building envelope of the case study house

Components Area (m2)

Walls above ground surface 170

Walls and floor below ground surface 134

Ceiling 93

Windows 13.4

Doors 8.8

Table 4

Thermal resistance of different components of the building envelope

Components Thermal resistance (m2K/W)

House Quebec law

Walls above ground surface 3.52 3.4

Walls in contact with ground 1.03 2.2

Floors in contact with ground 1.0 2.2

Attic-type roof 7.0 5.3

Table 5

Number of hours of occurrence of each hourly temperature bin for

Montreal, from CGCM2 A2x scenario

Temperature bin (1C) 1961–1990 2040–2069

–26 7 0

–25 14 0

�24 17 0

�23 22 0

�22 27 0

�21 32 9

�20 20 11

�19 33 15

�18 35 36

�17 54 24

�16 73 42

�15 79 34

�14 97 41

�13 144 64

�12 143 73

�11 158 107

�10 147 117

�9 148 137

�8 152 182

�7 173 173

�6 211 186

�5 221 213

�4 246 213

�3 221 233

�2 223 227

�1 197 251

0 203 228

1 207 242

2 186 232

3 185 203

4 176 207

5 155 214

6 143 164

7 139 171

8 122 135

9 117 139

10 100 123

11 77 110

12 72 88

13 59 78

14 49 68

15 42 60

16 34 49

17 32 40

18 28 34

19 13 31

20 20 24

21 13 17

22 4 13

23 7 9

24 1 6

25 5 4

26 1 4

27 2 3

28 2 2

29 0 2

R. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310306

new houses built in Montreal area. The natural airinfiltration rate in the attic is estimated at 0.3 ach. Thehouse is divided into two heated thermal zones: zone no. 1on the ground floor and zone no. 2 in the basement,and two unheated zones: zone no. 3 (garage) and zoneno. 4 (attic).

The electric baseboard heaters operate between October1 and April 30 to maintain the indoor air temperature ofthe ground floor space at 20 1C. There is a mechanicalventilation system that supplies 0.35 ach in the livingspaces. A heat recovery ventilator with an averageefficiency of 0.6 is used for preheating the cold ventilationair.

4.2. Weather data

The monthly values presented in Table 1 for eachanalysis period are used as input to the pre-processor ofweather data, called Type 54, which is a componentavailable in the TRNSYS environment. This moduleconverts the monthly average values into hourly valuesrequired by the computer model (TRNSYS, 2006; Knightet al., 1991). The number of hours of occurrence of eachtemperature bin, for both present and future climates,is extracted from the hourly outdoor air temperatures(Table 5). The number of occurrences of each dailytemperature bin is also extracted and presented in Table 6.

5. Analysis of results from synthetic data

Since the utility bills are the main source of historicaldata about the energy performance of houses, the proposedmethod should lead to accurate estimates using monthly orbi-monthly data. For the purpose of comparison, however,this section presents results that are obtained by using(i) the hourly data (Ei, T0,i), (ii) the daily data, and

(iii) the monthly data. Moreover, the results are comparedwith those from the detailed thermal simulation withTRNSYS.

ARTICLE IN PRESS

Table 6

Number of occurrences of each daily temperature bin for Montreal, from

CGCM2 A2x scenario

Temperature bin (1C) 1961–1990 2040–2069

�23 1 0

�22 0 0

�21 1 0

�20 1 0

�19 1 0

�18 2 1

�17 2 0

�16 2 1

�15 3 2

�14 4 1

�13 4 0

�12 4 4

�11 7 4

�10 5 3

�9 7 5

�8 7 8

�7 10 5

�6 7 9

�5 8 10

�4 15 8

�3 6 11

�2 12 15

�1 11 5

0 10 15

1 6 14

2 9 9

3 12 9

4 8 15

5 10 3

6 6 12

7 5 8

8 5 6

9 5 5

10 2 4

11 7 5

12 1 2

13 3 6

14 1 1

15 0 3

16 1 1

17 0 0

18 1 1

19 0 0

20 0 1

R. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310 307

In order to simplify the presentation, the energysignature is calculated directly for the present climate.Fig. 1 shows the hourly synthetic data over the heatingseason of the present climate along with the heating energysignature. The spread of points is due to the influence ofother parameters (e.g., solar radiation, thermal mass) onthe heating energy use, which is considered by the detailedsimulation. The energy signature is based only on theoutdoor air temperature.

Table 7 presents the coefficients of the hourly heatingenergy signature for both present and future climate, whichare extracted from synthetic data. The Student t-distribu-tion (Sheskin, 2004) is used to test the hypothesis that there

is no significant statistical difference between the slope ofthe energy signature for present climate and the slope forfuture climate, apresent ¼ afuture. The t-statistic is calculatedas follows:

t ¼ðapresent � afutureÞ

Sp

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1=Pn1

i ðT01;i � m01Þ2þ 1=

Pn2i ðT02;i � m02Þ

2q ¼ 0:24

(7)

with

S2p ¼

S21ðn1 � 2Þ þ S2

2ðn2 � 2Þ

ðn1 � 2Þ þ ðn2 � 2Þ, (8)

where m01 ¼ �1.95 1C is the average hourly air temperatureduring the heating season of the present climate;m02 ¼ 0.532 1C is the average hourly air temperature duringthe heating season of the future climate; S1 ¼ 9.18 1C is thestandard deviation of outdoor air temperature T01;S2 ¼ 8.48 1C is the standard deviation of outdoor airtemperature T02; n1 ¼ n2 ¼ 5088 is the number of hourlydata.The reference t-value is estimated in terms of the number

of degrees of freedom n ¼ n1+n2–4 ¼ 10,172 and level ofsignificance a ¼ 0.05: tn ¼ 1.645 (Sheskin, 2004). Sincetotn, the hypothesis apresent ¼ afuture is confirmed withthe risk of 0.05. Therefore, both heating energy signatures,for the present and future climates, have identicalsensitivity to outdoor air temperature. The energy signa-ture developed for the present climate can be used for thefuture climate.The annual heating energy use of the case study

house is estimated by TRNSYS at 10,252 kWh/year(54.0 kWh/m2yr) for the present climate, and 9012 kWh/year (47.4 kWh/m2yr) for the future climate, which gives areduction of 12.1%. The proposed method used withhourly data predicts that the climate change would reduceby 2040–2069 the heating energy use by 12.6% comparedwith 1961–1990 period (Table 8). Therefore, the differencebetween the proposed method and TRNSYS predictions ofthe climate impact on the heating energy use is of 0.5%.The calculations presented above with the proposed

method are repeated by using: (1) the daily total energy useand the corresponding daily average outdoor air tempera-ture, and (2) the daily average energy use of each period i,which is obtained from the monthly or bimonthly valuedivided by number of days of that period i, along with thecorresponding average outdoor air temperature.The coefficients of the heating energy signature are

presented in Table 9. The Student t-distribution is also usedwith the daily and monthly data to test the hypothesis thatthere is no significant statistical difference between theslope of the heating energy signature of present climate andthe slope for the future climate (Table 10). Since totn, theheating energy signatures for the present and futureclimates have identical sensitivity to outdoor air tempera-ture for a sampling rate of 1 day and 1 month, respectively.

ARTICLE IN PRESS

-2.00

-1.00

0.00

1.00

2.00

3.00

4.00

5.00

6.00

-30.00 -20.00 -10.00 0.00 10.00 30.00

To

E (

kW

h)

20.00

Fig. 1. Graphical representation of the hourly heating energy use and energy signature of the case study house vs. outdoor air temperature for the present

climate.

Table 7

Coefficients a and b of the heating energy signature of the house based on

hourly data, from CGCM2 A2x scenario

Period a (kWh/1C) b (kWh) R2

1961–1990 �0.102 1.816 0.72

2040–2069 �0.097 1.823 0.70

Table 8

Changes of the heating energy use as predicted by the proposed method,

from CGCM2 A2x scenario

Period Proposed method (kWh)

Hourly data

1961–1990 10,251

2040–2069 8963

Impact of climate change �12.6%

Daily total data

1961–1990 10,212

2040–2069 9018

Impact of climate change �11.7%

Daily average from monthly data

1961–1990 10,252

2040–2069 8981

Impact of climate change �13.1%

Table 9

Coefficients a and b of the heating energy signature based on daily and

monthly data

Period a (kWh/1Cday) b (kWh/day) R2

Daily total data

1961–1990 �2.30 43.81 0.87

2040–2069 �2.18 43.68 0.85

Daily average from monthly data

1961–1990 �2.57 43.38 0.94

2040–2069 �2.47 43.84 0.92

Table 10

Summary of reference t-value and t-statistic

Hourly data Daily data Daily average from

monthly data

Reference t-value

tn

1.645 1.645 1.812

t-Statistic 0.24 1.16 0.166

R. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310308

The difference between the predictions of the proposedmethod (12.6%, 11.7%, and 13.1%, respectively) andthose from detailed modeling with TRNSYS (12.1%) is lessthan 1% regardless of the duration of sampling period(Table 8). Therefore the method can be used with dataextracted from the utility bills. The additional cost andtime needed for collecting more detailed hourly or dailydata is not justified by the small increase in accuracy ofresults.

6. Sample of existing houses

This section presents the application of the proposedmethod on a sample of 11 existing houses. The heatingcontribution to the total annual energy use is between 60%and 81%. It is important to mention that none of selectedhouses have a mechanical ventilation system. Only housesusing natural gas or heating oil for heating purposes areincluded in this section. The information was collected byZmeureanu et al. (1999) from utility bills that rely on actualreadings. Table 11 presents as an example data extractedfrom the utility bills of house no. 44 that uses heating oil,while the corresponding heating energy signature ispresented in Fig. 2. The daily average heating energy useEh,i, in equivalent-kWh per day, of each billing period iscalculated as follows:

Eh;i ¼ AHV=ð3:6 NodaysÞ, (9)

ARTICLE IN PRESSR. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310 309

where A is the total amount of heating oil, in liters,delivered to the house at the end of each billing period, asindicated in the utility bills; HV is the heating value,considered equal to 39.0MJ/l; Nodays is the number ofdays of each billing period. In the case of natural gas and

0.0

50.0

100.0

150.0

200.0

250.0

-20 -10 0 10 20

T outdoor (C)

Daily a

vera

ge e

nerg

y u

se

(kW

h/d

ay)

Fig. 2. Graphical representation of daily average heating energy use and

energy signature of house no. 44 vs. daily average outdoor air

temperature.

Table 11

Data extracted from the heating oil bills of house no. 44

Month Day Year Heating oil use (litres)

12 9 1992 –

1 7 1993 235

2 4 1993 264

3 4 1993 282

4 9 1993 184

11 4 1993 363

12 14 1993 303

1 11 1994 443

2 7 1994 479

3 7 1994 415

4 19 1994 374

Table 12

Impact of climate change (1961–1990 vs. 2040–2069) on a sample of existing h

House Year Heated floor area

(m2)

Coefficients of heating energy

a (kWh/1Cday) b (kWh

10 1978 246 �4.70 137.51

12 1955 170 �7.72 109.75

17 1965 203 �7.04 151.54

18 1939 102 �7.74 139.37

22 1920 223 �9.46 150.79

24 1973 205 �6.50 82.62

26 1945 198 �8.00 175.21

28 1985 225 �7.34 116.82

38 1965 218 �6.06 131.67

40 1920 200 �6.52 139.58

44 1935 230 �5.11 82.11

A is the total volume of natural gas delivered to the house,in m3, at the end of each billing period; HV ¼ 37.3MJ/m3.The heating energy signature of house no. 44 has the

following form:

E ¼ �5:11T0 þ 82:11, (10)

where E is the daily average heating energy use, inkWh/day; T0 is the daily average outdoor air temperature;the weather-dependent coefficient a ¼ �5.11kWh/(1Cday),and the weather-independent coefficient b ¼ 82.11kWh/day.The coefficients a and b are developed by the least-squaresmethod, applied to the number of available pairs (Eh,i, T0).The resulting coefficient of determination R2 is equalto 0.86.Table 12 presents the heating energy signature of each

house based on the daily average values. All those existinghouses are more sensitive to climate, due to less effectivethermal insulation and heating system, than the case studyhouse presented in Section 5. The coefficient has a valuesbetween �4.70 kWh/1Cday and �9.46 kWh/1Cday for thesample of 11 existing houses, while for the case study housea ¼ �2.57 kWh/1Cday. When the daily average outdoorair temperature is 0 1C, the heating energy use of the casestudy house is 43.38 kWh/day, while the energy use ofsample existing houses is between 82.0 kWh/day (housesno. 24 and 44) and 175.0 kWh/day (house no. 26). The yearof construction is not always a reliable indicator of theheating energy use because some houses were renovatedand the set point temperature for heating was not kept atthe same level in all houses. For instance, the predictedannual heating energy use for the future climate of2040–2069 is 281.1 kWh/m2 yr for house no. 18, built in1939, and only 73.2 kWh/m2 yr for house no. 44, built in1935. These two houses have also the extreme annualenergy use for heating.The potential reduction of the heating energy use, due to

the climate change, of the sample of 11 existing housesvaries between 7.9% for house no. 10 and nearly 17% forhouse no. 24 (Table 12).

ouses, using monthly temperatures data from CGCM2 A2x scenario

signature Heating energy consumption (kWh/

m2 yr)

Impact of climate

change (%)

/day) 1961–1990 2040–2069

126.4 116.3 �7.9

155.6 131.7 �15.3

172.6 154.3 �10.6

321.0 281.1 �12.4

160.8 138.6 �13.9

98.5 81.9 �16.9

204.3 183.0 �10.4

123.5 106.4 �13.9

139.5 124.9 �10.5

161.4 144.3 �10.6

84.8 73.2 �13.8

ARTICLE IN PRESSR. Zmeureanu, G. Renaud / Energy Policy 36 (2008) 303–310310

7. Conclusions

The proposed method can be used to predict thepotential change in heating energy use by using historicaldata available in utility bills. The heating energy signaturecan be developed from monthly or bi-monthly utility bills,which rely on readings, and the corresponding averageoutdoor air temperature. If each utility bill covers a longerperiod, e.g., 3–4 months, the energy signature cannot beaccurately derived only from a few points, and thereforethe proposed method might not lead to accurate estimates.If the utility bills are based on energy use estimates ratherthan on readings, the derived energy signature is mean-ingless for the purpose of proposed method.

The method assumes that the energy signature of thehouse does not change in time between the present andfuture climatic conditions, which are predicted by theCCCMA.

The comparison between the predictions of the proposedmethod, using synthetic data, and those from the detailedbuilding energy analysis environment (TRNSYS) shows adifference less than 1%. It is expected that the annualenergy use for heating the case study house would bereduced in 2040–2069 by 11–13.1% depending on thesampling rate of heating energy use data. The reduction ofannual heating energy use of the sample of 11 existinghouses varies between 7.9% and 16.9%.

Future work should involve (i) the collection of datafrom a large sample of existing houses, eventually throughthe collaboration of utility companies, and (ii) theestimation of potential impact of climate change on theheating energy use in terms of the year of construction andaccording to different emission scenarios.

References

Belzer, D.B., Scott, M.J., Sands, R.D., 1995. Climate change impacts on

US commercial building energy consumption: an analysis using sample

survey data. Energy Sources 18, 177–201.

Caunesil, C., Callies, M., Zmeureanu, R., Roux, J-J., 2004. Comparison

of two building energy-related standards: MNECCB (Canada) and

RT-2000 (France). In: CIB World Congress.

CCCMA, 2007. Canadian Centre for Climate Modelling and Analysis

/http://www.cccma.ec.gc.ca/models/models.shtmlS (last accessed

June 2007).

CICS, 2007. Canadian Institute for Climate Studies /www.cics.uvic.caS(last accessed June 2007).

Deeble, V.C., Probert, S.D., 1986. Straight-line correlations for annual

energy-consumption predictions. Applied Energy 25, 23–39.

Fels, M.F., 1986. PRISM: an introduction. Energy and Buildings 9, 5–18.

Gaterell, M.R., McEvoy, M.E., 2005. The impact of climate change

uncertainties on the performance of energy efficiency measures applied

to dwellings. Energy and Buildings 37, 982–995.

IPCC, 2001. Climate Change 2001: The Scientific Basis. Cambridge

University Press, Cambridge.

Jacobsen, F.R., 1985. Energy signature and energy monitoring in building

energy management systems. In: Proceeding of CLIMA 2000 World

Congress, vol. 3: Energy Management, pp. 25–31.

Knight, K.M., Klein, S.A., Duffie, J.A., 1991. A methodology for the

synthesis of hourly weather data. Solar Energy 46 (2), 109–120.

Levermore, G., Chow, D.H., 2004. Climate change assessment and

building performance. In: CIB World Building Congress 2004:

Building for the Future, Toronto.

Lyberg, M.D., 1987. Source Book for Energy Auditors, vol. 1.

International Energy Agency, Anex XI-Energy Auditing, Swedish

Council for Building Research, Stockolm, pp. 38–42.

Mirasgedis, S., Sarafidis, Y., Georgopoulou, E., Kotroni, V., Lagouvar-

dos, K., Lalas, D.P., 2007. Modeling framework for estimating

impacts of climate change on electricity demand at regional level: case

of Greece. Energy Conversion and Management 48, 1737–1750.

Nakicenovic, N., Swart, R. (Eds,), 2000. Special Report on Emissions

Scenarios, Intergovernmental Panel on Climate Change.

Ouranos, 2004. In: DesJarlais, C., Bourque, A., Decoste, R., Demers, C.,

Deschamps, P., Lam, K.-H. (Eds.), Adapting to Climate Change.

Ouranos, Montreal, ISBN2-923292-01-4.

Quebec, 2005. Regulation respecting energy conservation in new build-

ings. Last amendment: 24 September 1992, Editeur official du Quebec.

Sheskin, D.J., 2004. Handbook of Parametric and Nonparametric Statistical

Procedures, third ed. Chapman & Hall/CRC, Boca Raton, FL.

Thatcher, M., 2006. Modelling changes to electricity demand load

duration curves as a consequence of predicted climate change for

Australia. Energy 32, 1647–1659.

TRNSYS, 2006. A Transient System Simulation Program. Version 16.

Solar Energy Laboratory, University of Wisconsin-Madison, WI,

USA.

Zmeureanu, R., 1990. Assessment of the energy savings due to the

building retrofit. Building and Environment 25 (2), 95–103.

Zmeureanu, R., 1992. A new method for evaluating the normalized energy

consumption in office buildings. Energy 17 (3), 235–246.

Zmeureanu, R., Fazio, P., DePani, S., Calla, R., 1999. Development of an

energy rating system for existing houses. Energy and Buildings 29,

107–119.